Tags:
, view all tags

-- PeterKammel - 15 Feb 2007

My most recent report on entrance muon detector inefficiencies dates from 5/9/06:

http://weak0.physics.berkeley.edu/weakint/research/muons/private/tbanks_dir/Documents/UCB-Run8-updated-results-v2.pdf

If I had the luxury of time, I would run high-stat MC simulations again with the following improvements:

1. Reduce the deadtime from 12 ns -> 7 ns, and eliminate the 40% muPC1 "recovery fraction." Of course, this would probably have almost no effect on the fit results, because both 7 and 12 ns are far below the fit start time of 100 ns, and 12*.6 ~ 7 anyways. 2. Reduce the time-independent inefficiency from 34 ppm -> 8 ppm, in light of the results presented in http://weak0.physics.berkeley.edu/weakint/research/muons/private/tbanks_dir/TeleConf/2006Jun27/2006Jun27_updated.html The earlier inefficiency estimate of 34 ppm was based on drift histograms which included muon TPC stop candidates before fiducial cuts were applied, so I think the 34 ppm contains a lot of background junk. This means that my existing MC t-indep estimates are a very conservative 4x too high. 3. The MC code now has the capability to look at a wider breakdown of lifetime spectra background and signal contributions (pure and pure+wallstop decay signals, TIIE ABG, TDIE ABG, etc.), which would enable the disentanglgin of contributions which Peter was asking for. I have run this improved MC configuration only once with Run8 (2e9) stats: http://weak0.physics.berkeley.edu/weakint/research/muons/private/tbanks_dir/TeleConf/2006Dec12/2006Dec12.html

Finally, concerning the hypothesis that the mysterious tail in the right-hand accidental background of the drift distribution is due to deadtimes: I first raised this possibility by voice only (10/4/05 teleconf?), in the very early days of my studies and before I had time to consider the MC results more carefully. By the next teleconference I had already discarded the idea:

http://weak0.physics.berkeley.edu/weakint/research/muons/private/tbanks_dir/TeleConf/2005Oct11/2005Oct11.html

Figures 5 and 6 are the plots of interest, and I conclude that

"... it appears that a simple 30 ns deadtime isn't sufficient to replicate what we see in the Run8 drift distribution (Figure 4); you need a longer [mus] deadtime in order to see some accidental BG "spill" over into the gap between the drift distribution and pileup background."

For Peter K.: You were asking about the evidence of deadtime in the drift plots, which I present in Figures 11 and 12 of my 5/9/06 report. My October 11, 2005 worklog page above gives more explanation which you might find helpful:

"The 30 ns deadtime certainly produces a nonuniform accidental BG--it basically generates a miniature copy of the drift signal distribution. I would guess that the t-dependent accidental BG is a convolution of the drift peak with a square pulse of width=deadtime."

This is what you are seeing the in the red histogram bins underneath the main coincidence peak in Figures 11 and 12.

Regards, Tom

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r1 - 2007-02-15 - PeterKammel
 
  • Edit
  • Attach
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback